On 10/25/21 3:12 PM, Ilya Leoshkevich wrote:
__BYTE_ORDER is supposed to be defined by a libc, and __BYTE_ORDER__ -
by a compiler. bpf_core_read.h checks __BYTE_ORDER == __LITTLE_ENDIAN,
which is true if neither are defined, leading to incorrect behavior on
big-endian hosts if libc headers are not included, which is often the
case.
Instead of changing just this particular location, replace all
occurrences of __BYTE_ORDER with __BYTE_ORDER__ in libbpf code for
consistency.
ACK, that is definitely broken as is - we had similar issue back then with the
bpf_{htons,ntohs}() helpers, details: 78a5a93c1eeb ("bpf, tests: fix endianness
selection").
The bpf_core_read.h change I would split out as a separate commit along with a
proper Fixes tag so it could potentially be cherry-picked easier (since the
remainder in here is a cleanup for consistency and not used out of the BPF prog
where this issue exists).
Signed-off-by: Ilya Leoshkevich <iii@xxxxxxxxxxxxx>
---
tools/lib/bpf/bpf_core_read.h | 2 +-
tools/lib/bpf/btf.c | 4 ++--
tools/lib/bpf/btf_dump.c | 8 ++++----
tools/lib/bpf/libbpf.c | 4 ++--
tools/lib/bpf/linker.c | 12 ++++++------
tools/lib/bpf/relo_core.c | 2 +-
6 files changed, 16 insertions(+), 16 deletions(-)
Thanks,
Daniel
